Between 25%-50% of the severely mentally ill (SMI) have extensive substance abuse problems. Which substance abuse treatments work best for the SMI and how to provide them have received little systematic study. This study compares the cost-effectiveness of three substance abuse treatment approaches for the severely mentally ill (SKI), in terms of (a) direct and indirect societal costs, and (b) treatment outcomes in four major categories: (1) alcohol and drug use (2) level of psychiatric symptomatology, (3) social functioning, and (4) reported life satisfaction. At study entry SKI clients (N-260) have already been randomly assigned to and are receiving case management and outpatient mental health services from 9 treatment in Santa Clara County. In this study, we will augment this primary mental health care with substance abuse treatment. The three substance abuse treatment interventions to be tested differ in clinical orientation: (1) is based on a coping and relapse prevention skills modal of behavior change; (2) is based on a community reinforcement model; (3) is based on a social recovery/AA-RA model, and is the approach currently being used by the service systems ("control condition"). Three treatment teams will provide services based on each treatment model. Client outcomes and societal costs will be monitored for two years after study entry. The relative cost-effectiveness of these three intervention models will constitute the main comparisons. We will identify which types of clients benefit most from the interventions by incorporating initial severity of substance abuse, mental health status, sex, age, ethnicity, and other client characteristics into the regression analyses. The effects of environmental differences on client outcome will also be examined by incorporating variables such as living arrangement and participation in self-help groups in the comparative analyses. This study will contribute significantly to the field in several ways: (a) The treatment interventions to be studied are outpatient treatment programs monitored over a two year period. (b) This research examines treatment provided to the large client population seen in public agencies and tests models of integrating substance abuse services with primary mental health treatment for the SMI. (e) This study will empirically test the application of treatment models that have been successful with primary drug and alcohol abusers to the SKI with secondary substance abuse problems and represents a pioneering effort in the application of economic analyses, i.e., the study of cost-effectiveness of these treatments